public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH] Fix simplify_const_unary_operation (PR debug/48204)
@ 2011-03-23 20:11 Jakub Jelinek
  2011-03-24 10:02 ` Richard Guenther
  0 siblings, 1 reply; 2+ messages in thread
From: Jakub Jelinek @ 2011-03-23 20:11 UTC (permalink / raw)
  To: gcc-patches

Hi!

simplify_const_unary_operation assumes it doesn't have to change anything
in REAL_VALUE_TYPE for FLOAT_EXTEND, which is wrong when FLOAT_EXTENDing
e.g. from SFmode to DDmode or SDmode to DFmode, where the number
needs to be converted from binary to decimal or vice versa.

Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux,
ok for trunk/4.6.1?

2011-03-23  Jakub Jelinek  <jakub@redhat.com>

	PR debug/48204
	* simplify-rtx.c (simplify_const_unary_operation): Call
	real_convert when changing mode class with FLOAT_EXTEND.

	* gcc.dg/dfp/pr48204.c: New test.

--- gcc/simplify-rtx.c.jj	2011-03-21 13:00:09.000000000 +0100
+++ gcc/simplify-rtx.c	2011-03-23 16:39:13.000000000 +0100
@@ -1526,7 +1526,8 @@ simplify_const_unary_operation (enum rtx
     }
 
   else if (GET_CODE (op) == CONST_DOUBLE
-	   && SCALAR_FLOAT_MODE_P (mode))
+	   && SCALAR_FLOAT_MODE_P (mode)
+	   && SCALAR_FLOAT_MODE_P (GET_MODE (op)))
     {
       REAL_VALUE_TYPE d, t;
       REAL_VALUE_FROM_CONST_DOUBLE (d, op);
@@ -1549,7 +1550,10 @@ simplify_const_unary_operation (enum rtx
 	  d = real_value_truncate (mode, d);
 	  break;
 	case FLOAT_EXTEND:
-	  /* All this does is change the mode.  */
+	  /* All this does is change the mode, unless changing
+	     mode class.  */
+	  if (GET_MODE_CLASS (mode) != GET_MODE_CLASS (GET_MODE (op)))
+	    real_convert (&d, mode, &d);
 	  break;
 	case FIX:
 	  real_arithmetic (&d, FIX_TRUNC_EXPR, &d, NULL);
--- gcc/testsuite/gcc.dg/dfp/pr48204.c.jj	2011-03-23 16:44:09.000000000 +0100
+++ gcc/testsuite/gcc.dg/dfp/pr48204.c	2011-03-23 16:43:35.000000000 +0100
@@ -0,0 +1,10 @@
+/* PR debug/48204 */
+/* { dg-do compile } */
+/* { dg-options "-O -fno-tree-ccp -fno-tree-dominator-opts -fno-tree-fre -g" } */
+
+void
+foo (void)
+{
+  float cf = 3.0f;
+  _Decimal64 d64 = cf;
+}

	Jakub

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH] Fix simplify_const_unary_operation (PR debug/48204)
  2011-03-23 20:11 [PATCH] Fix simplify_const_unary_operation (PR debug/48204) Jakub Jelinek
@ 2011-03-24 10:02 ` Richard Guenther
  0 siblings, 0 replies; 2+ messages in thread
From: Richard Guenther @ 2011-03-24 10:02 UTC (permalink / raw)
  To: Jakub Jelinek; +Cc: gcc-patches

On Wed, Mar 23, 2011 at 9:11 PM, Jakub Jelinek <jakub@redhat.com> wrote:
> Hi!
>
> simplify_const_unary_operation assumes it doesn't have to change anything
> in REAL_VALUE_TYPE for FLOAT_EXTEND, which is wrong when FLOAT_EXTENDing
> e.g. from SFmode to DDmode or SDmode to DFmode, where the number
> needs to be converted from binary to decimal or vice versa.
>
> Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux,
> ok for trunk/4.6.1?

Ok.

Thanks,
Richard.

> 2011-03-23  Jakub Jelinek  <jakub@redhat.com>
>
>        PR debug/48204
>        * simplify-rtx.c (simplify_const_unary_operation): Call
>        real_convert when changing mode class with FLOAT_EXTEND.
>
>        * gcc.dg/dfp/pr48204.c: New test.
>
> --- gcc/simplify-rtx.c.jj       2011-03-21 13:00:09.000000000 +0100
> +++ gcc/simplify-rtx.c  2011-03-23 16:39:13.000000000 +0100
> @@ -1526,7 +1526,8 @@ simplify_const_unary_operation (enum rtx
>     }
>
>   else if (GET_CODE (op) == CONST_DOUBLE
> -          && SCALAR_FLOAT_MODE_P (mode))
> +          && SCALAR_FLOAT_MODE_P (mode)
> +          && SCALAR_FLOAT_MODE_P (GET_MODE (op)))
>     {
>       REAL_VALUE_TYPE d, t;
>       REAL_VALUE_FROM_CONST_DOUBLE (d, op);
> @@ -1549,7 +1550,10 @@ simplify_const_unary_operation (enum rtx
>          d = real_value_truncate (mode, d);
>          break;
>        case FLOAT_EXTEND:
> -         /* All this does is change the mode.  */
> +         /* All this does is change the mode, unless changing
> +            mode class.  */
> +         if (GET_MODE_CLASS (mode) != GET_MODE_CLASS (GET_MODE (op)))
> +           real_convert (&d, mode, &d);
>          break;
>        case FIX:
>          real_arithmetic (&d, FIX_TRUNC_EXPR, &d, NULL);
> --- gcc/testsuite/gcc.dg/dfp/pr48204.c.jj       2011-03-23 16:44:09.000000000 +0100
> +++ gcc/testsuite/gcc.dg/dfp/pr48204.c  2011-03-23 16:43:35.000000000 +0100
> @@ -0,0 +1,10 @@
> +/* PR debug/48204 */
> +/* { dg-do compile } */
> +/* { dg-options "-O -fno-tree-ccp -fno-tree-dominator-opts -fno-tree-fre -g" } */
> +
> +void
> +foo (void)
> +{
> +  float cf = 3.0f;
> +  _Decimal64 d64 = cf;
> +}
>
>        Jakub
>

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2011-03-24 10:02 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-03-23 20:11 [PATCH] Fix simplify_const_unary_operation (PR debug/48204) Jakub Jelinek
2011-03-24 10:02 ` Richard Guenther

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).